Research Designs, Essay Example

In order to understand the advantages and disadvantages of different research study designs, this paper will take a look at two groups of 1200 Middle Eastern women (600 women from Egypt; 600 women from Saudi Arabia) sampled from rural areas.[1] In these experiments, the dependent variable being evaluated is attitude towards gender egalitarianism; the independent variables are the women’s nature of religious level (measured in a binary “orthodox” or “liberal” binary).  The scheduled intervention is a “women’s rights seminar” that will give a more nuanced (read: liberal) appraisal of women’s function in Islamic society.

For the pre-test/post-test control group design, the group of Egyptian women will be broken into two groups: 1) One group of 200 Egyptian women will be exposed to a “gender egalitarian” one hour long seminar talking about equal gender roles in an Islamic context with discussion after; one group of 200 Egyptian women will simply be put into a room for one hour meeting asking about their views towards life with no discussion afterwards.  Both of the groups will be given a pre-test and post-test assessing their views on gender roles in Islamic society. The same design, with the Saudi women split into two equal groups of 200, will also be performed. The pre-test/ post-test structure is typically used because there is a control group whose progress can be measured against the intervention- that is, measure the true impact of the intervention.

For the Solomon-Four-Group Design, a different group of 400 Egyptian and 400 Saudi women will each be broken into four different groups consisting of 100 women in each group: 1) The first group will be assigned to the pre-test regarding gender egalitarian issues, the intervention, and the post-test; 2) The second group will be exposed to the intervention but only with a post-test; 3) The third group will take the pre- and post-test on gender equality without the intervention; 4) The fourth group will just take the post-test.

The third test would be the “post-test” only group.  In these tests, the last group of 400 Egyptian and Saudi women will be broken down into two groups of 200.  The first group will be exposed to the seminar on gender equality and a post-test on views regarding gender equality; the second group will not be exposed to the intervention but will take a post-test regarding views on gender equality. The post-test experimental method is typically used if the investigator is worried about how the influence of testing may make subjects sensitive to the intervention, ultimately confounding results.

Overall, there are seven threats to internal validity (validity of putative relationship between dependent and independent variable) to analyze.  A majority of the threats to internal validity are mitigated through the use of randomized allocation and the use of a control group in experiments.  For example, the internal validity threats of maturation, testing, and regression are mitigated using a control group.  Maturation, the phenomenon by which changes in groups are a function of development (increased capacity) rather than the intervention, will be equally noticeable in both groups.  Regression to the mean is a similar threat: a control group will track the regression to the mean (mirroring a similar trend in the intervention group) indicating that the intervention is not the primary factor behind different results. Testing, the process by which participants may change answers due to multiple testing, is controlled for by a control group and is a main reason for the Solomon four group and post-test design. Randomizing subjects solves the internal validity problem of selection, in which the differing characteristics of subjects in different groups is confounding the analysis.

In the three different research designs, the internal validity threat of history is mitigated by the randomization of timing and experimenters giving the test.  The internal  validity threat of mortality is controlled for by having a placebo discussion session for the control group of Middle Eastern women.  Instrumentation is controlled for by giving the same test- there is little discretion left up to the experimenters for personal decision that could introduce bias.  Regarding the threat of attrition, the “intent-to-treat” analysis will be administered in the experiments leaving in data points even if they ultimately choose to exit the experiment.  The diffusion threat to internal validity will be controlled by keeping the women in different hotels and arranging their sessions individually and not as an entire group based on their random assignment before the experiment begins.  Finally, sequencing events should not be a foremost concern in any of these studies because there are no crossover designs.

[1] Note: defined as cities smaller than 25,000- at least 100 miles away from a city that is larger than 100,000).  Rural women are chosen because they are less likely to have been influenced to more liberal interpretations of Islam.