Survey Research - Science method

Hi everyone,

I am currently in the process of reviewing feedback responses from my pilot survey (mixed methods with a mixture of closed and open ended questions).

One of the feedback that I have received from many of the respondents was they would like to choose more options in the question that I have asked e.g. Describe the type of service delivery that you provide (please pick up to three).

For this type of question, I don't think forced choice or check all that apply would be appropriate and I am trying to prevent people from satisficing answers for CATA. Most clinicians (targeted population) work with a variety of clients and my goal here is to get them to pick top three that they spent the most time in. I am not looking for them to rank based on my research question but what I am looking for is to see certain relationship e.g. respondents working in the education department seemed to be more confident in x. So basically in my case, all the options provided are of equal standing and what I am just looking for is for them to pick 3.

As mentioned, many of the respondents want to pick more than 3 and I am not sure what number is appropriate. I have tried looking for guidance by reading research articles related to survey designs and nothing relevant seemed to come up.

Basically my question is: does anyone know of any research evidence behind these multiple selection questions (why only three can be chosen) and if there is some formula to decide how many options the respondents should choose based on the number of options provided (e.g. if there are 6 options on the response stem choose only 2, if 12 choose 4 etc? ). If there is an expert in this field that has done specific research in this - please let me know of their names and I will look them up.

Thanks in advance.

Gen

A number of people have asked on ResearchGate about acceptable response rates and others have asked about using nonprobability sampling, perhaps without knowing that these issues are highly related.  Some ask how many more observations should be requested over the sample size they think they need, implicitly assuming that every observation is at random, with no selection bias, one case easily substituting for another.   

This is also related to two different ways of 'approaching' inference: (1) the probability-of-selection-based/design-based approach, and (2) the model-based/prediction-based approach, where "prediction" means estimation for a random variable, not forecasting. 

Many may not have heard much about the model-based approach.  For that, I suggest the following reference:

Royall(1992), "The model based (prediction) approach to finite population sampling theory." (A reference list is found below, at the end.) 

Most people may have heard of random sampling, and especially simple random sampling where selection probabilities are all the same, but many may not be familiar with the fact that all estimation and accuracy assessments would then be based on the probabilities of selection being known and consistently applied.  You can't take just any sample and treat it as if it were a probability sample.  Nonresponse is therefore more than a problem of replacing missing data with some other data without attention to "representativeness."  Missing data may be replaced by imputation, or by weighting or reweighting the sample data to completely account for the population, but results may be degraded too much if this is not applied with caution.  Imputation may be accomplished various ways, such as trying to match characteristics of importance between the nonrespondent and a new respondent (a method which I believe has been used by the US Bureau of the Census), or, my favorite, by regression, a method that easily lends itself to variance estimation, though variance in probability sampling is technically different.  Weighting can be adjusted by grouping or regrouping members of the population, or just recalculation with a changed number, but grouping needs to be done carefully. 

Recently work has been done which uses covariates for either modeling or for forming pseudo-weights for quasi-random sampling, to deal with nonprobability sampling.  For reference, see Elliott and Valliant(2017), "Inference for Nonprobability Samples," and Valliant(2019), "Comparing Alternatives for Estimation from Nonprobability Samples."  

Thus, methods used for handling nonresponse, and methods used to deal with nonprobability samples are basically the same.  Missing data are either imputed, possibly using regression, which is basically also the model-based approach to sampling, working to use an appropriate model for each situation, with TSE (total survey error) in mind, or weighting is done, which attempts to cover the population with appropriate representation, which is mostly a design-based approach. 

If I am using it properly, the proverb "Everything old is new again," seems to fit here if you note that in Brewer(2014), "Three controversies in the history of survey sampling," Ken Brewer showed that we have been all these routes before, leading him to have believed in a combined approach.  If Ken were alive and active today, I suspect that he might see things going a little differently than he may have hoped in that the probability-of-selection-based aspect is not maintaining as much traction as I think he would have liked.  This, even though he first introduced 'modern' survey statistics to the model-based approach in a paper in 1963.  Today it appears that there are many cases where probability sampling may not be practical/feasible.  On the bright side, I have to say that I do not find it a particularly strong argument that your sample would give you the 'right' answer if you did it infinitely many times when you are doing it once, assuming no measurement error of any kind, and no bias of any kind, so relative standard error estimates there are of great interest, just as relative standard error estimates are important when using a prediction-based approach, and the estimated variance is the estimated variance of the prediction error associated with a predicted total, with model misspecification as a concern.  In a probability sample, if you miss an important stratum of the population when doing say a simple random sample because you don't know the population well, you could greatly over- or underestimate a mean or total.  If you have predictor data on the population, you will know the population better.  (Thus, some combine the two approaches: see Brewer(2002) and Särndal, Swensson, and Wretman(1992).) 

..........         

So, does anyone have other thoughts on this and/or examples to share for this discussion: Comparison of Nonresponse in Probability Sampling with Nonprobability Sampling?    

..........         

Thank you.

References:

Brewer, K.R.W.(2002), Combined Survey Sampling Inference: Weighing Basu's Elephants, Arnold: London and Oxford University Press

Brewer, K.R.W.(2014), "Three controversies in the history of survey sampling," Survey Methodology, Dec 2013 -  Ken Brewer -   Waksberg Award: 

https://www150.statcan.gc.ca/n1/pub/12-001-x/2013002/article/11883-eng.htm

Elliott, M.R., and Valliant, R.(2017), "Inference for Nonprobability Samples," Statistical Science, 32(2):249-264,

https://www.researchgate.net/publication/316867475_Inference_for_Nonprobability_Samples, where the paper is found at

https://projecteuclid.org/journals/statistical-science/volume-32/issue-2/Inference-for-Nonprobability-Samples/10.1214/16-STS598.full (Project Euclis, Open Access). 

Royall, R.M.(1992), "The model based (prediction) approach to finite population sampling theory," Institute of Mathematical Statistics Lecture Notes - Monograph Series, Volume 17, pp. 225-240.   Information is found at

https://www.researchgate.net/publication/254206607_The_model_based_prediction_approach_to_finite_population_sampling_theory, but not the paper. 

The paper is available under Project Euclid, open access: 

https://www.doi.org/10.1214/lnms/1215458849

Särndal, C.-E., Swensson, B., and Wretman, J.(1992), Model Assisted Survey Sampling, Springer-Verlang

Valliant, R.(2019), "Comparing Alternatives for Estimation from Nonprobability Samples," Journal of Survey Statistics and Methodology, Volume 8, Issue 2, April 2020, Pages 231–263, preprint at 

https://www.researchgate.net/publication/337950671_Comparing_Alternatives_for_Estimation_from_Nonprobability_Samples 

At the US Energy Information Administration (EIA), for various establishment surveys, Official Statistics have been generated using model-based ratio estimation, particularly the model-based classical ratio estimator.  Other uses of ratios have been considered at the EIA and elsewhere as well.  Please see

 https://www.researchgate.net/publication/261508465_Use_of_Ratios_for_Estimation_of_Official_Statistics_at_a_Statistical_Agency

At the bottom of page 19 there it says "... on page 104 of Brewer(2002) [Ken Brewer's book on combining design-based and model-based inferences, published under Arnold], he states that 'The classical ratio estimator … is a very simple case of a cosmetically calibrated estimator.'" 

Here I would like to hear of any and all uses made of design-based or model-based ratio or regression estimation, including calibration, for any sample surveys, but especially establishment surveys used for official statistics. 

Examples of the use of design-based methods, model-based methods, and model-assisted design-based methods are all invited. (How much actual use is the GREG getting, for example?)  This is just to see what applications are being made.  It may be a good repository of such information for future reference.

Thank you.  -  Cheers. 

Research methodologists have identified serious problems with the use of "control variables" (aka nuissance variables, covariates), especially in survey research. Among the problems are uninterpretable parameter estimates, erroneous inferences, irreplicable results, and other barriers to scientific progress. For the sake of discussion, I propose that we stop using control variables altogether. Instead, any variable in a study should be treated like the others. If it is important enough to include in the research, it is important enough to be included in your theory or model. Thoughts for or against this proposition?

Kindly Provide your suggestions its an important part of my research work.

I have been trying to understand research paradigms (neo- positivism, interpretivism/social construction and critical realism) for a few days now, and I've been reading a number of resources, primarily Blaikie and priest's Social research: Paradigms in action (2017), and Tracy's Qualitative research method. In Blaikie and priest, they say that paradigms are used at the level of explanation, but when I read Tracy's work, I get the impression that paradigms come into play at the level of description as well. These various descriptions creates more confusion for me. At what level of research do these paradigms come into play?

In addition to this, I have been reading many articles that does no seem to follow the descriptions of the paradigms strictly. Are there some researches that don't usually follow?

In light of these two, do you think that survey research follows these paradigms?

Looking forward to reading your views and thought.

In survey research, it is advised that the response rate should be high to avoid self-selection bias. What methods can be used to assess if the data is affected by biases resulting from low response rate,

I supposed to collect data from 384 respondents. But, I only get 230 complete responses in return. In this case, my response rate is only 60%. Is it acceptable?

What are the components that should best describe the organization of a study related to a medical survey proposal?

can I use a combination of a vignette (for one variable i.e. dependent variable) with a self-report survey questionnaire (for all other variables IVs, Mediators, and moderators)? if I can what types of analysis and software for that analysis I may use? if I can't what should I do? (scale development is not a good solution, neither scale for survey research is used nor available in previous research for that Dependent variable). I mean can I use a vignette for one variable with a self-report scale for all other variables in combination (it is somehow a mix of experimental and self-report methodology).

Is the sample size for CFA the same like that of EFA in educational survey research?

Hello Researchers,

Can anyone recommend a good and professional survey data-collecting platform in China?

Hi. I am a bachelor degree student and currently conducting a survey research about Knowledge, Attitude and Purchasing behaviour towards chicken egg, and willingness to pay for chicken egg. May I know what is the independent and dependent variable for this research? I'm still new with this type of research method and confused which variable is which. I really appreciate your help and thank you.

Dear colleagues,

I am preparing an instrument to better understand factors influencing the research agenda setting of researchers working in academic and non-academic settings. I would like to ask your collaboration to make it better, by completing it and leave comments at the end of the questionnaire on how to improve it.  Many thanks in advance.

Survey link: https://www.surveymonkey.com/r/WNTJNTC

p.s.: the survey is voluntary and anonymous, and I would appreciate it if you could circulate it among colleagues that would be willing to contribute and help. Many thanks.

As internet is, now-a-days, a useful tool for data collection and research, issues also go hand in hand with ethical approval and informed consent. As participants are, by default, give their consents as they comply with the survey, still the question of ethical approval remains unanswered. While there is no need to obtain ethical approval in secondary data and in some similar circumstances, what about primary research of internet based survey?

This anonymous survey is open to all UK and Middle East academics, researchers, postgraduate students, and professionals. It takes 10 minutes to complete. At the end of the survey you will be offered the opportunity to fill in your details on a separate online form, in case you wish to be considered for the prize draw. To participate, please click on the link below. You are welcome to share the link with your professional and/or social network too.

This is a survey for a Master’s thesis and your support is greatly appreciated. The title of the study is ‘‘The role of leadership self-efficacy (LSE) in developing academic and professional leaders’’. You can find more information in the Participant Information Sheet, which is available with the survey.

I am going to conduct a survey among the experts who are working in power plant construction projects in my country. So far I know, construction of 30 mega projects are going on. The targeted experts are 7-categories, for example, contractors, sub-contractors, vendors, project director (PD), project manager (PM), site engineer, and consulting engineer (or consultant). The other variables are project size in terms of power generation capacity, budget, project location, experts' experience (year), and academic qualification. Please suggest me in a precise way to save my time? I am now in a critical moment. I have a presentation just after couple of weeks. Thank you for your patience and time. 

I'm conducting a comparative research of omnichannel experience. Kindly ask to take part in this survey https://impresaluiss.qualtrics.com/jfe/form/SV_7WZOTN1rENKA4EC

Highly appreciate your help!

Dear academy colleagues,

I'm looking for a truly comprehensive resource for teaching graduate students the elements of conducting robust survey research, including proper survey development, validation, distribution, confidentiality, data security, collation, statistical analysis, interpretation, and sense-making. I've seen elements of these in resources here and there, but not complete, and usually not written in a way that is accessible to a graduate student.

Do you have recommendations about a single resource or a progression of resources that really help a student get from zero to fairly strong (obviously with practice and some mentoring)?

I'm a bit confuse how to measure students' learning. I have used word 'learning' in my research topic and now I'm struck what kind of questionnaire or any other tool would be used to measure students' learning not their performance or achievement. I need a clarification about the term 'Learning"

I am trying to perform the cell-weighting procedure on SPSS, but I am not familiar with how this is done. I understand cell-weighting in theory but I need to apply it through SPSS. Assume that I have the actual population distributions.

We are conducting survey research on COVID-19 misinformation among students. How can we validate our survey for it to produce valid and reliable data? Also, what statistical tools can we use to measure the extent of misinformation in a community? Thank you.

I conducted an experimental study to examine the effect of the different communication methods {two experimental groups (method A and method B) with the control group}. All participants were randomly assigned to one of the experiment conditions (i.e. communication method A, B, or control) after exposure to each method respondents were asked to indicate their agreement or disagreement for five statements about the influence of the communication method using a 4-point Likert scale (strongly disagree, disagree, agree, strongly agree) with a not-applicable option.

I coded “not applicable” responses in two different ways based on experimental conditions. First, if a respondent is exposed to method A and method B and chosen “not-appliable” as a response for five statement items I coded response as Zero. Treating “not applicable” as zero was appropriate in this situation because the respondent put themselves to the lower end by indicating no effect of treatments (Huggins-Manley et al., 2018; Welch, 2013). Second, if a respondent in the control group chooses “not applicable” for any of five statements it was coded four. The “not applicable” response for the control condition describes their true situation (i.e. communication method did not influence them) so I gave a score (4) for the control group. Using a predetermined value for not-applicable responses reduces variance. So I would like to seek any advice on how to code “not-applicable” responses appropriately for such a situation.

Huggins-Manley, A. C., Algina, J., & Zhou, S. (2018). Models for semiordered data to address Not applicable responses in scale measurement. Structural Equation Modeling: A Multidisciplinary Journal, 25(2), 230-243. https://doi.org/10.1080/10705511.2017.1376586

Welch, W. W. (2013). Developing and analyzing a scale to measure the impact of the advanced technological education program. https://www.evalu-ate.org/resources/doc-2013-dev-analyzing/

can quantitative research i.e. survey researches be exploratory in nature?

How can I determine the sample size for my survey if I have two sample frames?

Do I calculate it separately for each?

The two designs seem to have some common characteristics as both of them correlate variables. Therefore, how could one differentiate between them?

For example: Would using a 7 point scale of 0 - 3 in steps of 0.5 give you different results to using a 7 point scale of 0 - 6 in steps of 1? I'm aware that verbal labels are likely to be better but I'm interested in the possible differences between purely numerical scales that use 0.5 or 1 increments.

Thanks.

I developed my own survey based on previous themes from qualitative data. Pairs of questions were extracted from subscales from previous questionnaires in the field, and then adapted to fit the survey context. I have now completed data collection and I ran a CFA on the 'a-priori' factors and questions that were developed, and the model fit wasn't great!

I went back to the data and conducted an EFA to see what factors did work together, and the reliability, plus overall model fit when doing CFA was much, much better. The factors extracted during EFA weren't that far from the original themes, except for a couple of questions being moved around.

Therefore, my question is - is this a done thing? As this was a data-driven survey, would it be acceptable to run EFA and go by this factor structure to continue with the rest of my stats? Or should I just stick with the original 'a-priori' factor structure and deal with the consequences?

Thanks!

I plan to conduct an online survey via Survey Monkey. It is a quantitative study that will measure perceived stress and I plan to use purposive sampling. No correlations, just a simple survey research design on a specific sample. I endeavor to use the Statistical Package for Social Sciences (SPSS) I am not too sure what will be the best form of statistical analysis if my target population will be 100? Thank you in advance for your advice.

I will do single group pre post test in evaluating training. In that case, what strategy i should follow between experimental or survey research?

Using a questionnaire as a data collection instrument in a survey, do a researcher need to formulate hypothesis?

Hello,

I'm not home in survey research among children (between 6-12 years old). I'm looking for scales for children which measure stress, coping, relationships with peers etc. Any suggestions?

Kind regards,

Filip

What i mean to say is measurement like strongly agree, agree neutral, disagree and strongly disagree need to be analysed using ordinal logistic regression. Because there is no quantification of how many times of "agree" is ""strongly agree"? similarly how may times of disagree is strongly disagree. Can any on explain

Anybody pls explain and give a model research report

I hope to conduct a series of interviews/questionnaire surveys to collect information regarding urban flood management and the use of software tools for the same.

Fundamentally, decision-makers, flood modellers, general public and software modellers/developers are in my expected audience.

Could you please suggest what personal information should be considered when weighing them?

My assumptions are as follow;

1. Decision Makers: The age, level of education, years of service, the level in the organization, no of participations/decision makings in actual flood management activities

2. Flood modellers: educational status (MSc/PhD etc), years of experience, no of participations/decision makings in actual flood management activities

3. Software developers: years of experience, no of contributions in actual flood management software development and the role he/she played

4. General Public: The Age, the level of flood-affected to the person, educational level, experience with floods

We conducted a survey about knowledge of a specific topic (measured as a score) and got some complete responses and some partial responses (demographics only). We think that non-response may be an indicator of lack of knowledge. How can I analyze the data to confirm that?

We thought to compare the demographics between full responders and partial responders and see whether the significantly different variables are the same variables that predict knowledge in the full response group. But we are looking for a better analysis approach that can combine the two outcomes in the same analysis (response to the survey and knowledge about the topic). Any advice?

I am carrying out a statistical analysis of my collected data through interviews. I would like to interpret the RSD in terms of the level of precision (i.e., Excellent, acceptable, unacceptable). what happens if I got 30% RSD or 60% RSD?! What does this mean in terms of precision? I would appreciate if you support your answers with references.

Hi all,

I've got some questions for my thesis. I do a cross-sectional quantitative research (OLS regressions) based on an archival survey conducted between 2015 and 2017. The survey was not administered by me. I hope you can help.

1) What is more preferable, dropping control variables to have a relatively large sample size or remain controls and have a low sample size? (Especially in my case the difference is big n=86 with controls Firm Age and Firm Size and without n=120) OLS assumptions not violated in both situations.

2) Do I have to mention inter-reliability (e.g. Cronbach’s Alpha) among items of the survey? Or can I assume the items as inter-reliable since the administrators did not mention it.

Regards,

Luuk

Dear all

I am currently doing some literature survey in research on diabetic. I want some literature evidence to make some decision. Please share if you have any literature evidence.

Thanking you

Umaramani.M

I plan to carry out a survey research that will model a certain variable with other pre-specified variables. As this is not an intervention study, I am not obliged to register the study to such a database as clinicaltrials.gov. However, I plan to do that in order to increase transparency of my research. Particularly, I would like to make all the tested variables prospectively revealed in order to make the statistical modelling more valid.

Do you consider my plan correct? Would you give me any further tips on how to increase transparency of my study?

I have studied telephone interviews that are conducted for marketing or survey research (Barriball et al. 1996; Carr & Worth 2001) but in case of semi-structured interviews, is it appropriate to conduct through any medium (Skype, Viber, etc) on video or audio? Or onsite is only recommended and mandatory?

I have studied telephone interviews that are conducted for marketing or survey research (Barriball et al. 1996; Carr & Worth 2001) but in case of semi-structured interviews, is it appropriate to conduct through any medium (Skype, Viber, etc) on video or audio? Or onsite is only recommended and mandatory?

I am using multi research design for my research. i.e. Survey, experimental and correlation. It is not only Survey research. My respondents are student teachers of Pune University. Population is almost 250000. Please guide me how much sample should I take for survey?

Please, I need responses.

Thank you.

I am conducting an empirical research and wondering how to gather around 2000-3000 respondents email ids, considering the response rate of 10 - 15%.

Does empirical researchers who are regularly conducting this type of research would like to share their experience. As it would be highly appreciable.

Common method bias is apparent in cross-sectional survey research. Are there significant ways to minimise the common method bias in cross-sectional survey research? Please provide me with the sources.

What is 'survey' in research? Type of research, type of research design, method of research or method of data collection?

Questionnaires and Checklists seem to be the most widely instruments used for data collection in survey studies among undergraduate Students.

Social desirability bias is a significant challenge in Survey-based studies. Are there any significant ways to minimise the social desirability bias in Survey research, particularly in the field of consumer behaviour and insights?

I am looking at evaluations which ask participants of a program to rate different aspects of the program from 1-5. The evaluation was sent to all participants and around 20% responded (~650 people).  I'm worried about non-response bias, but not sure how to test whether my estimates are significantly biased since my data is ordinal. Any thoughts? 

Many thanks!

The Slovin's Formula is quite popularly use in my country for determining the sample size for a survey research, especially in undergraduate thesis in education and social sciences, may be because it is easy to use and the computation is based almost solely on the population size. The Slovin's Formula is given as follows: n = N/(1+Ne²), where n is the sample size, N is the population size and e is the margin of error to be decided by the researcher. However, its misuse is now also a popular subject of research here in my country and students are usually discourage to use the formula even though the reasons behind are not clear enough to them. Perhaps it will helpful if we could know who really is Slovin and what were the bases of his formula.

Hi, I am reading a book on research methodology and wondering whether Survey Research method is inductive or deductive. Can survey research be in the form of qualitative data?

Thank you

I am doing survey research of customer right now so that I want to identify the minimum percentage that can be stated customer satisfy towards the service and quality of company. May I guess is it between 70 % and 90 %. I need the exact percentage about this. May anyone can give me assistance to answer my question. Thanks a bunch.

Is there a possibility that the data are valid and realiable if you use a self -made qustionaire.

Requesting references

Hello everyone, I am currently working on my thesis where I am analyzing if some factors have impact on purchasing behavior. 15 of the questions in my survey were were Likert Scale (1-5) and I would like to use them as independent variable. Total money spend would be dependent variable and I would also like to use income and gender and some other stuff as independent variable. What statistical analysis would you recommend me to analyze my data? I converted my Likert scale data into factor data with 5 levels and now I am lost.

Can I use OLS or would you recommend something else?

Any help would be really appreciated

As you are doubtless aware, paper-based survey has been known as one of the most common methods for gathering data relevant to people's behavior (either revealed preferences or stated preferences). I wanna make sure how much can we rely on new methods like Internet (Web)-based survey instead of traditional paper-based survey? In particular, my research's scope is related to travel behavior analysis. My research' sample should cover all socioeconomic groups and almost all geographical areas in a city.

I would be happy if somebody shared with me his/her opinion or the valid references.

Thanks in advance

Design-based classical ratio estimation uses a ratio, R, which corresponds to a regression coefficient (slope) whose estimate implicitly assumes a regression weight of 1/x.  Thus, as can be seen in Särndal, CE, Swensson, B. and Wretman, J. (1992), Model Assisted Survey Sampling, Springer-Verlang, page 254, the most efficient probability of selection design would be unequal probability sampling, where we would use probability proportional to the square root of x for sample selection. 

So why use simple random sampling for design-based classical ratio estimation?  Is this only explained by momentum from historical use?  For certain applications, might it, under some circumstances, be more robust in some way???  This does not appear to conform to a reasonable data or variance structure.

I am conducting survey research on MTurk that involves participants reading workplace sexual harassment scenarios that include potentially distressing depictions of sexual assault/violence. I want to design the survey in Qualtrics such that participants: (1) must spend a certain amount of time reading each scenario before advancing to dependent measures and next scenario (i.e., placing timer on block) and (2) may, at any point, advance directly to the end of the study to receive MTurk completion code if they wish to terminate participation for whatever reason. Is there a way achieve both of these requirements, or will I have to abandon (1) and allow participants to advance freely to the end of the survey by skipping questions? This is more of Qualtrics survey flow logic question.

More generally: how do researchers handle mid-study termination of participation and payment in Qualtrics survey research on MTurk?

Whether research survey is mandatory or not?

I am currently creating a survey around communication within the feasibility assessment process of construction projects. And if possible I would like my respondents to all come from the industry so that my research knowledge is reliable. How can I make sure I channel it to the correct people in the correct industry?

Dear colleagues,

I will be really thankful if you share your experience , research, pre-prints on the impact of GDPR on planning and executing the surveys. How you achieve the consent for data protection? Whaat is the respondent reaction? What is the response rate after GDPR? We expect some decrease, for now more visible in online surveys than in face-to-face.

I'm exploring innovative ways of conducting research with teens and younger children. We are looking to measure positive youth development, and normal surveys are not very engaging for this demographic, so I'm curious if anyone has explored any creative ways - e.g., gamification, turning survey into a fun quiz, using apps, etc. Looking for quantitative suggestions at this stage.

Thank you!

Can I use purposive sampling in a quantitative survey research?

In Applied Survey Sampling, by Blair and Blair, Sage Publications, 2014, on page 175, they note the common use of "credibility intervals," by researchers using nonprobability samples and Bayesian modeling. They note that the American Association for Public Opinion Research (AAPOR) issued a caution to people using these credibility intervals, as not being something the public should rely on "in the same way" as a margin of sampling error. Attached is the AAPOR statement from 2012 in which they caution heavily regarding the use of such nonprobability opinion polls, as the Bayesian models have assumptions which will be of varying quality. However, they also state that "...even the best design [probability sample] cannot compensate for serious nonparticipation by the public." Thus, for much of the data in a probability sample, we have to use models or some method to estimate for missing data, and when the nonresponse rate is high, do we really actually have a valid probability sample anymore?

Thus the emphasis would be on total survey error. There is sampling error, and then there is nonsampling error. We have nonresponse and that can make reliance on a model better overall.

If that is the case, then why do many survey statisticians insist on probability samples for highly skewed establishment surveys with continuous data, when excellent regressor data are available? Often sampling the largest few establishments will provide very high 'coverage' for any given important variable of interest. That is, most of the estimated total would already be observed. The remainder might be considered as if it were missing data from a census. But these missing data, if generally close to the origin in a scatterplot of y vs regressor x, should have little prediction error considering the heteroscedastic nature of such data. With the relatively high measurement error often experienced with small establishments, long experience with energy data has shown one will often predict values for y for small x more accurately than one could observe such small y's. Further, this is done using the econometric concept of a "variance of a prediction error," and Bayesian model assumptions are not introduced.

It is important not to lump nonprobability sampling having good regressor data with other nonprobability sampling. For official statistics, an agency will often collect census surveys, and have more frequently collected samples of the same variables of interest (the same attributes), or a subset of them. Often the best regressor data for the samples are from such a census.

Finally, many years of use in publishing official statistics on energy data have shown this methodology - far less radical than the use of "credibility intervals" for polling - to have performed very well for establishment surveys. It does not appear reasonable to argue with such massive and long-term success. The second link attached is a short paper on the use of this methodology. The third is a study of the variance and bias, and the fourth shows a simple example, using real data, as to how effective a quasi-cutoff sample with a model-based classical ratio estimator can be.

Since the 1940s, for good reason in many cases, probability sampling has been the "gold standard" for most sample surveys. However, models are used heavily in other areas of statistics, and even for survey statistics "model-assisted design-based" methods have unquestionably often greatly improved probability sample results. But strictly model-based sampling and estimation does have a niche in establishment surveys, though it has met with resistance. One should not dismiss this without trying it. It especially seems rather odd if anyone were to consider Bayesian "credibility intervals" for election polls, but not quasi-cutoff sampling with model-based estimation, as defined by going through the second link below.

Your comments? 

My topic is Relationship between transformational leadership style and affective organizational commitment: a study based on functional level employees .

Also i would like to ask in questionnaire to determine the transformational leadrship behaviour can i write "My supervisor" Communicate a convincing vision for the future. or should i replace it with my leader?

Can I use purposive sampling in a quantitative survey research?

I am currently working on my master’s thesis about Sustainable Marketing in Construction Industry, at Istanbul Technical University. My research aims to introduce marketing as a tool for sustainable construction business development for the creation of the sustainable built environment. It should only take 10 minutes to complete. All responses will remain anonymous.

Please find below the link to the survey:

http://veti.itu.edu.tr/form/bsertyesilisik/sustainablemarketinginconstructionindustry

Thanks in advance for your support

hi everyone expert in statistics,

my sample size is 16 respondents in a survey research. So , to measure correlation, what type of analysis is most suitable??

I have seen several references to "impure heteroscedasticity" online as heteroscedasticity caused by omitted variable bias.  However, I once saw an Internet reference, as I recall, which reminds me of a phenomenon where data that should be modeled separately are modeled together, causing an appearance of increased heteroscedasticity.  I think there was a youtube video.  That seems like another example of "impure" heteroscedasticity to me. Think of a simple linear regression, say with zero intercept, where the slope, b, for one group/subpopulation/population is slightly larger than another, but those two populations are erroneously modeled together, with a compromise b. The increase in variance of y for larger cases of x would be at least partially due to this modeling problem.  (I'm not sure that "model specification error" covers this case where one model is used instead of the two - or more - models needed.)

I have not found that reference online again.  Has anyone seen it? 

I am interested in any reference to heteroscedasticity mimicry.  I'd like to include such a reference in the background/introduction to a paper on analysis of heteroscedasticity which, in contrast, is only from the error structure for an appropriate model, with attention to unequal 'size' members of a population.  This would then delineate what my paper is about, in contrast to 'heteroscedasticity' caused by other factors. 

Thank you. 

I'd like to know what insights were generated and what the implications were for research or commercial outfits.

I have conducted pre-test surveys of few hundred people through non-probability sampling. Now can I create a sampling frame work with the details of those pre-test survey respondents and than select the random sample for my final survey from this ?

If allowed what would be that sampling technique called? Would it be probability? Also can we use these results to generalise it on health food population in my city (target population) or only limited to sample frame work. Please guide as this would be very useful. Thanks.

I am looking for ways to address validity and reliability of the data collected with this technique.

Stack Overflow advertises several official (https://stackoverflow.com/help/how-to-ask) and de facto standard guidelines (http://tinyurl.com/stack-hints) for writing good questions -- that is, questions that have greater chances to be resolved.

We built a statistical model to empirically validate this question-writing netiquette. The last step we're missing in our research is: What does the SO community think of these recommendations?

If you ever used Stack Overflow before, please help us find out by taking this very short survey: https://goo.gl/EzS3eN

Thanks for your contribution! 

The time to completion for the survey is about 5 minutes only (before June 21, 2017).

This is a purely academic research project with no commercial interests. We will openly publish the results so everyone can benefit from them, but will anonymize everything before doing so; your responses will be handled confidentially. Please, note that you are not obligated to participate in the survey. If at some point during the survey you want to stop, you are free to do so without any negative consequences. Incomplete survey data will not be used.

This survey is anonymous. The record of your survey responses does not contain any identifying information about you, unless a specific survey question explicitly asked for it.

If I pick up data from a survey for only 10% and randomly generate the rest of 90% from an application. (based on the 10%)  Will this work?  I am in IS discipline.

I think many people do simulate things in other domains too.