Social Science Research Barriers
Take Steps to Ensure Reliability and Validity in Studies
In order to conduct social science research, one must have good measurement. Without it, the researcher can't argue that the conclusions of the research are meaningful.
In an uncontrolled, imperfect world, researchers are already fighting an uphill battle as they try to introduce order or control to their studies. Unsound measurements are one more strike against the researcher who seeks to eliminate background noise in his research. Some areas of scientific inquiry can make a reader challenge the researcher’s assertions.
Self-Report Measures
Much social science research relies on the reports, memories, and experiences of human participants. Researchers need to know what people actually did or thought, but this is difficult to ascertain. For example, if a researcher is comparing someone’s attitude about an issue five years ago to that person’s attitude about the same issue today, the person’s memory could be cloudy. Individuals who’ve had a major attitude shift may also answer inaccurately to relieve cognitive dissonance.
Another problem with self-reports is the phenomenon of giving socially desirable answers. Research subjects might say what they think the researcher wants them to say, or they might answer in a way that reflects social norms. Researchers can use control checks, such as asking a question in several different ways, to increase the validity of self-report claims.
Cause-and-Effect Statements
Researchers sometimes make cause-and-effect statements when they aren’t appropriate. A researcher shouldn’t make causal arguments following descriptive research in the social sciences. However, not even experimental research methods always yield a causal argument. For example, researchers may make claims of causality when another variable confounds the argument. For instance, a researcher may argue that shy students have a greater tendency to enroll in early classes, because students in the 8 a.m. classes don’t engage in class discussion. However, the researcher failed to account for the variable of fatigue, which may keep early risers from speaking up.
Generalizing Research Findings
When researchers generalize the findings of their research to large populations, the reader must closely consider the external validity of the study. One reason researchers take so much care in their sampling methods is that they want to state that what’s true for the sample is also true for the population.
The reader must consider some questions regarding the ability to generalize study results. Is the sample typical and representative? The media might try to convince us that young people today have a propensity for mass murder by focusing on a few sensational cases, but these cases don’t accurately reflect the population. In addition, can others replicate the study? The more times researchers replicate their findings with different groups of people, the more comfortable the reader can be that the findings are applicable to large populations.
Sound Statistics
Researchers can skew quantitative reports based on samples just like any other part of a study’s design. For instance, if a researcher conducting a study on computer-mediated communication draws heavily from statistics published in a study 10 years ago, one might question the value of the findings. In general, statistics are applicable only to the time in which the researcher generated them.
Sampling is another area fraught with the potential for errors. If some researchers want to study spousal abuse, and the sample they draw is limited to subscribers of Good Housekeeping, it’s unlikely that the findings will be applicable to other groups.
The procedures and measures must also stand up to peer scrutiny. If a social science researcher wants to study what makes employees steal office supplies, and the supervisor takes part in collecting the data, the findings would be questionable.
Finally, the presentation of the data must accurately represent the statistical findings. Researchers must take care to compose their charts or graphs carefully so that they don’t suggest something the data doesn’t support.
Learn more about reducing researcher bias and increasing the ability to generalize study results to large populations.
Sources:
Reinard, J. (1998). Introduction to Communication Research, 2nd Ed. McGraw-Hill: Boston.
Verdugo, E.D. (1998). Practical Problems in Research Methods. Pyrczak Publishing: Los Angeles.
