Higher Code Commits Predicts Very Slightly Lower Daily Step Count for Population

Daily Step Count was generally 7% higher than average after a total of 5 event of Code Commits over the previous 7 days.

Aggregated data from 1 study participants suggests with a MEDIUM degree of confidence (p=0.0967, 95% CI -1337.531 to 1337.421) that Code Commits has a very weakly negative predictive relationship (R=-0.055) with Daily Step Count.

The highest quartile of Daily Step Count measurements were observed following an average 5.97 event Code Commits per day.

The lowest quartile of Daily Step Count measurements were observed following an average 5.59 event of Code Commits per day.

After an onset delay of 0 seconds, Daily Step Count is typically 4% higher than average over the 7 days following around 5.59 event Code Commits.

The objective of this study is to determine the nature of the relationship (if any) between Code Commits and Daily Step Count. Additionally, we attempt to determine the Code Commits values most likely to produce optimal Daily Step Count values.

Manual Recording Option

Create a reminder for Code Commits here and record it daily by enabling notifications or using the reminder inbox here .

Manual Recording Option

Create a reminder for Steps here and record it daily by enabling notifications or using the reminder inbox here .

This study is based on data donated by 1 participants. Thus, the study design is equivalent to the aggregation of 1 separate n=1 observational natural experiments.

Code Commits Pre-Processing

Code Commits measurement values below 0 event were assumed erroneous and removed. No maximum allowed measurement value was defined for Code Commits. It was assumed that any gaps in Code Commits data were unrecorded 0 event measurement values.

Steps Pre-Processing

Steps measurement values below 1 count were assumed erroneous and removed. No maximum allowed measurement value was defined for Steps. No missing data filling value was defined for Steps so any gaps in data were just not analyzed instead of assuming zero values for those times.

Predictive Analytics

It was assumed that 0 seconds would pass before a change in Code Commits would produce an observable change in Daily Step Count.

It was assumed that Code Commits could produce an observable change in Daily Step Count for as much as 7 days after the stimulus event.

Using a two-tailed t-test with alpha = 0.05, it was determined that the change in Daily Step Count is statistically significant at 95% confidence interval.

After treatment, a 9.9% decrease (-1370 count) from the mean baseline 12700 count was observed. The relative standard deviation at baseline was 49.8%. The observed change was 0.216881 times the standard deviation.

A common rule of thumb considers a change greater than twice the baseline standard deviation on two separate pre-post experiments may be considered significant. This occurrence would have only a 5% likelihood of resulting from random fluctuation (a p-value < 0.05).

Code Commits data was primarily collected using QuantiModo. QuantiModo allows you to easily track mood, symptoms, or any outcome you want to optimize in a fraction of a second. You can also import your data from over 30 other apps and devices. QuantiModo then analyzes your data to identify which hidden factors are most likely to be influencing your mood or symptoms.

Daily Step Count data was primarily collected using Fitbit. Fitbit makes activity tracking easy and automatic.

As with any human experiment, it was impossible to control for all potentially confounding variables. Correlation does not necessarily imply causation. We can never know for sure if one factor is definitely the cause of an outcome. However, lack of correlation definitely implies the lack of a causal relationship. Hence, we can with great confidence rule out non-existent relationships. For instance, if we discover no relationship between mood and an antidepressant this information is just as or even more valuable than the discovery that there is a relationship.

We can also take advantage of several characteristics of time series data from many subjects to infer the likelihood of a causal relationship if we do find a correlational relationship. The criteria for causation are a group of minimal conditions necessary to provide adequate evidence of a causal relationship between an incidence and a possible consequence.

Criteria For Causal Inference

Strength (A.K.A. Effect Size)

A small association does not mean that there is not a causal effect, though the larger the association, the more likely that it is causal. There is a very weakly negative (R = -0.055) relationship between Code Commits and Daily Step Count.

Consistency (A.K.A. Reproducibility)

Consistent findings observed by different persons in different places with different samples strengthens the likelihood of an effect. Furthermore, in accordance with the law of large numbers (LLN), the predictive power and accuracy of these results will continually grow over time. 248 paired data points were used in this analysis. Assuming that the relationship is merely coincidental, as the participant independently modifies their Code Commits values, the observed strength of the relationship will decline until it is below the threshold of significance. To it another way, in the case that we do find a spurious correlation, suggesting that banana intake improves mood for instance, one will likely increase their banana intake. Due to the fact that this correlation is spurious, it is unlikely that you will see a continued and persistent corresponding increase in mood. So over time, the spurious correlation will naturally dissipate.

Specificity

Causation is likely if a very specific population at a specific site and disease with no other likely explanation. The more specific an association between a factor and an effect is, the bigger the probability of a causal relationship.

Temporality

The effect has to occur after the cause (and if there is an expected delay between the cause and expected effect, then the effect must occur after that delay). The confidence in a causal relationship is bolstered by the fact that time-precedence was taken into account in all calculations.

Biological Gradient

Greater exposure should generally lead to greater incidence of the effect. However, in some cases, the mere presence of the factor can trigger the effect. In other cases, an inverse proportion is observed: greater exposure leads to lower incidence.

Plausibility

A plausible bio-chemical mechanism between cause and effect is critical. This is where human brains excel.

Based on our responses so far,

0 humans feel that there is a plausible mechanism of action for a relationship between Code Commits and Daily Step Count.

0 humans feel that any relationship observed between Code Commits and Daily Step Count is coincidental.

Coherence

Coherence between epidemiological and laboratory findings increases the likelihood of an effect. It will be very enlightening to aggregate this data with the data from other participants with similar genetic, diseasomic, environmentomic, and demographic profiles.

Experiment

All of human life can be considered a natural experiment. Occasionally, it is possible to appeal to experimental evidence.

Analogy

The effect of similar factors may be considered.

A plausible bio-chemical mechanism between cause and effect is critical. This is where human brains excel. Based on our responses so far, 0 humans feel that there is a plausible mechanism of action and 0 feel that any relationship observed between Code Commits and Daily Step Count is coincidental.