Psychology 2011 1997 Lab Week 2 Summary

These pages are demonstration pages of mixed mode deli very for an on-campus course. The curriculum and course structure has changed substantially for 1998, but these pages have been retained here as an example of a particular use of the Web in teaching.

Microsoft Excel Tutorial, Computer Attitude Survey Data and Issues in Survey Research

Introduction

The first part of the lab was devoted to teaching some basic skills in Microsoft Excel, which is a spreadsheet program that one can use to set up data files that ultimately might be used in a statistical package like SPSS (tutorials on SPSS are forthcoming). For some of you this part of the class may have been a simple refresher course, and hopefully you also got something out of the process. What I would like to highlight is the fact that at some point in your academic lives, you may be required to enter data obtained from surveys or questionnaires, in the way we entered just 6 subjects data in class. Given for example, 100 subjects completing a 30 item survey, with both positive and negative scored items, for 3 subscales (where these individual scores are wanted), you can begin to appreciate how long (and laborious) that process would be. Later in the year, you might have to collect data from research projects you have designed where the data collection method is surveys, and as such it will be you who is required to set up and enter a data file for future analyses. Human error in inputting data points is one of the most likely sources of data contamination. Care needs to be taken in entering data that has taken you x amount of time and energy to collect. Don't render any subsequent analysis invalid as a result of a sloppy data entry effort.

Excel Skills

In setting up the mini-data file it is hoped that you have now acquired the following skills;

you are confident in entering data and setting up a spreadsheet, including
- formatting cells (eg. bold, centering)
- inserting and deleting rows and columns,
- giving appropriate labels to factors in the experiment.
you understand the concept of sheets and know how to naming and move between sheets in an excel workbook.
you can complete the following basic editing commands; cut, paste, copy, moving rows/columns around a worksheet, the autofill function.
you are confident in using basic analysis tools in calculating the sum of a series of data points, obtain the means and SD, counting the number of data points in a sample (N).

A little about the short report

The skills listed above will be necessary when you come to play around with CAS excel data file (located on the Psych Notes system) in week 4. The lab in Week 4 has been set aside as a CAS data workshop. It is expected that before then you will have completed a literature search for references on the general area of computer attitude research, with the variables in our experiment (eg. sex, degree, computer and web usage (experience?)) in mind, so that by week 4 you have established an appropriate hypothesis for your short report. We do not expect in the short report an exhaustive literature review leading to your hypotheses. It is enough to read 2-3 references that you may cite in the brief introduction to establish the aims and hypotheses logically.

CAS data files

There are two CAS data files (Excel and SPSS) for you to play around with in probing data suitably for your hypotheses for the CAS short report. They are both located under the Psych Notes Finder, labelled appropriately, so you should have no trouble findng them.

Remember we have the following variables in our study;
Subject variables (our independent variables) - Age, Gender, Computer Use, Perceived Competency
Dependent Variables - CAS scores on Confidence, Anxiety, Liking, Overall score.

Issues in Survey Research

Survey Research is an extremely popular method of collecting data in psychological research. This is because it is relatively cheap and easy to implement, but also because in some situations it is in fact the best alternative for investigating a psychological phenomenon. In survey studies, researchers do not formally assign individuals to levels of the Independent Variable(s) (IV). Instead, it is subject characteristics (eg. age/sex/education) that are used to categorise the different levels of IV (often after the data have been collected. Comparisons between levels or between treatment conditions must always be made with the presumption that the groups are nonequivalent. Internal validity is therefore threatened, and causal inferences about the effects of IVs on DVs are difficult to make in survey designs.

Internal validity may not be a problem if the research question being asked is more concerned with how many / or what type of people act or believe in something. In this instance the researcher is not looking for cause-effect relationships.

With survey research it is crucial to measuring the psychological construct (eg. attitudes to computers) well to ensure that the collection of data is from a representative sample of individuals. A representative sample needs to be assured for external validity (the ability to generalise results to target population). There are a number of sampling issues that need consideration What type of sample you are using? The type of sampling could be a simple random sample, a stratified sample, or cluster sampling. An accurate estimation of sample size is also important for experimental validity and reliability.

The issues above are important for external validity - high external validity allows us to be confident in generalising the results to the population and to other settings that are of theoretical interest. The importance of external validity vary for different types of research, eg. national opinion polls v. class experiments (internal validity important in this case) The nature of desired generalisation may take different forms for different types of research. For example in surveying for opinion polls, or research on teaching methods, we have particularistic research goals. External validity amounts then to the ability to generalise the research results themselves from the studied sample to the target population. Sampling here is a crucial step.

Theoretically-driven projects investigating theoretically-hypothesised relationships, with no specific target population in mind, for example a study investigating a relationship between frustration and acts of aggressionp, have universalistic research goals. The consistency or inconsistency of the results obtained in the sample with the theoretically-based hypothesis is the key outcome, for inconsistency implies that the theory is inadequate and requires revision. The issuei here is to apply the theory itself outside the research context. We are not interested in applying the results to the target population but in building the theory for further interrogation.

Vague concepts like "self esteem" are difficult to manipulate into IVs, so we employ the best nonexperimental research design, recognising validity limitations and overcoming them as best as one can. Whenever we employ survey designs and have a goal of arguing for a causal hypothesis, we must attempt to eliminate alternative explanations for variation or changes in the DV. Appropriate statistical analysis is therefore critical so we can begin to rule out alternative explanations for "treatment" effects.

For many purposes, surveys and quasi-experimental designs are tools of choice rather than simply less desirable alternatives.

Finally in investigating the data derived from research studies, the participant's response tendency can tell us much about them as a person as the questions they are responding to, for example, what does it say about someone who indicates "neutral" to every question?. As researchers you also need to ensure that subject is responding consistently to the same sub-group of items - internal consistency, can be measured by obtaining a reliability coefficient, such as the coefficient alpha for each survey subscale. Often in journal articles on survey research you will see alpha's quoted as an indication of the surveys reliability.

Reference

A reference on this material will be provided very shortly. Stay tuned!!

Writing a draft Method Section

In writing the draft Method section please consider the following points;

Consider what experimental hypotheses you might test with the CAS.
Method section describes how you collect your data and how you propose to analyse it.
Describe subject population(s) you will be sampling.
Estimate numbers of subjects, age ranges, and any other important demographic information.
Method should be sufficiently detailed to allow replication of your experiment w/o irrelevant information.
Important to articulate precisely the form the data will take and how it will be analysed.

Checklist

The following things should be accomplished before week 3 labs

Attend library tutorial.

Acquire basic understanding of Microsoft Excel skills listed above.

Complete draft Method section for Week 3.

Begin to formulate research hypotheses for your short report.

Chris Hughes

Labs | PSY2011 Home Page | Psychology | Monash University