This seems to be yet another example of very dodgy science.
You take <20 women. They self-select for the type of contraceptive that they use. You then attempt to study the effect of contraceptive. Straight away you have a massive problem: the contraceptive is self-selected.
Think about that for a second. In the 21st century we have young women who are strippers, some of whom are voluntarily not using contraceptives and some who are not. How many reasons can you think of offhand as to why decisions leading to contraceptive choice might affect earnings all by itself. To look at the more obvious:
-
Women who were not using contraceptives were not in stable relationships. They were either celibate or engaging in causal sex and insisting on condom use. As such they lacked the emotional support structure of a committed relationship and were more prone to umpteen psychological effects.
-
Women who were not using contraceptives were physiologically infertile, either through surgery, illness or congenital defects. There are way too many ways for this to affect the results to go into here, but to give one obvious factor, most women get tubal ligations after they have already had children, so what effect does having children have all by itself?
-
Women who are not using contraceptives have more predictable menstrual cycles, less heavy menstrual flows or less severe PMS. Many women use birth control specifically to achieve a more predictable cycle or to alleviate either extreme PMS, especially cramps, or reduce menstrual flow. If this is the case then the women on contraceptives are already behind the 8 ball when it comes to earning.
Looking at the actual paper, the first thing that leaps out at me is how few only a single factor was evaluated.
For example, women not using contraceptives consistently earned more than women who were on all except 3 days a month. That should immediately set alarm bells ringing that the two groups were not selected form the same population. Yet there was not statistical evaluation done to determine if the populations were the same, only an evaluation to see whether phase had any effect on earnings. All that was analysed was whether earning varied between phases *within *groups. No analysis at all on whether it varied *between *groups.
Then we note that while the contraceptive group had a low point of just $75, it peaked at $250, a tripling of income over the cycle. Meanwhile the no-contraceptive group had a low point of $175 and a peak of $375, which is less than a doubling of income. So in fact the contraceptive group had a *greater *increase in earnings over their cycle. Yet no analysis of or discussion of this point is made at all. All that was examined was whether the difference was greater between phases. That is incredibly poor science considering that the studies is supposed to be looking at the effects of cycle phase on earning.
Then we look at the shapes of the curves. The contraceptive group has a nice regular curve, from a low immediately after onset on menstruation to a peak 2 1/2 weeks later followed by a gradual decline. The no-contraceptive group does indeed peak during the fertile cycle, but that peak starts to rise while menstruation is still occurring. Moreover there is a secondary peak in the infertile period in the week prior to menstruation, and that secondary peak is fully 2/3 as great as the putative fertile peak. Yet no analysis was done on the curve forms. No analysis of whether they are the same curve aside form the fertile peak (they clearly are not), no analysis of whether the fertile peak is significantly higher then the infertile peaks.
The only factor that the study looked at is whether contraception use resulted in a greater increase between phases within each group. It seems like a classic example of examining the data for the result that you want, and ignoring the results that are staring you in the face.
Consider me unimpressed. This would be an interesting submission to “Annals of Improbable Research”, but as a serious article, it’s third rate at best.