Many of these studies suffer from similar limitations and weaknesses, with the main obstacle being the difficulty in acquiring representative, random samples on a virtually invisible population. Many lesbian and gay parents are not open about their sexual orientation due to real fears of discrimination, homophobia, and threats of losing custody of their children. Those who do participate in this type of research are usually relatively open about their homosexuality and, therefore, may bias the research towards a particular group of gay and lesbian parents.
Because of the inevitable use of convenience samples, sample sizes are usually very small and the majority of the research participants end up looking quite homogeneous—e.g. white, middle-class, urban, and well-educated. Another pattern is the wide discrepancy between the number of studies conducted with children of gay fathers and those with lesbian mothers...
Another potential factor of importance is the possibility of social desirability bias when research subjects respond in ways that present themselves and their families in the most desirable light possible. Such a phenomenon does seem possible due to the desire of this population to offset and reverse negative images and discrimination. Consequently, the findings of these studies may be patterned by self-presentation bias.
According to a 2001 review of 21 studies by Stacey and Biblarz published in American Sociological Review: "[R]esearchers lack reliable data on the number and location of lesbigay parents with children in the general population, there are no studies of child development based on random, representative samples of such families. Most studies rely on small-scale, snowball and convenience samples drawn primarily from personal and community networks or agencies. Most research to date has been conducted on white lesbian mothers who are comparatively educated, mature, and reside in relatively progressive urban centers, most often in California or the Northeastern states."